Delhi NCR cmip aoi · delhi
lon 77.21000, lat 28.64000 · 1 km MGRS cell 43RGM1670
ingest/geocode.py: "Nominatim geocodes to a place, so precision is place-level (~3 km), never a street fix." The lon/lat above is a named-place point, not an address; the card describes its 1 km cell.
| This-week probability | 0.14 (0.08–0.25) · climatology baseline |
|---|---|
| Band | Medium |
| Top reasons | not computable — no mapped waterlogging hotspot lies in cell 43RGM1670 — of the 19 delhi hotspots, none is in this 1 km cell (nearest: Karol Bagh, 2.51 km away in cell 43RGM1371; its reasons describe that spot, not this cell) |
| Relative risk (within city) | 0.41 · above 76% of the 3245 boxes in this city — relative susceptibility within this city (0-1, quantile-banded), not a probability and not comparable between cities |
| Return period (RSK-06) | not computable — 4 usable annual maxima (< 5, RSK-06 STOP IF): years [2019, 2023, 2024, 2025] over a 7-year span |
| Depth, 1-in-10 yr | not computable — no depth layer exists: RSK-09's loss cards (site/public/data/loss_cards/) were never produced (deferred) and the HAND depth layer failed its own gate (HYD-08 STOP IF, EVALUATION.md) |
| Depth, 1-in-50 yr | not computable — no depth layer exists: RSK-09's loss cards (site/public/data/loss_cards/) were never produced (deferred) and the HAND depth layer failed its own gate (HYD-08 STOP IF, EVALUATION.md); and the record cannot support it anyway: no return period rarer than about 1-in-(n+1) years is computable from 4-5 dated events per AOI (RSK-02 STOP IF, EVALUATION.md 'Return-period frequency — record adequacy') |
| Depth, 1-in-100 yr | not computable — no depth layer exists: RSK-09's loss cards (site/public/data/loss_cards/) were never produced (deferred) and the HAND depth layer failed its own gate (HYD-08 STOP IF, EVALUATION.md); and the record cannot support it anyway: no return period rarer than about 1-in-(n+1) years is computable from 4-5 dated events per AOI (RSK-02 STOP IF, EVALUATION.md 'Return-period frequency — record adequacy') |
- This is a per-cell card, not a per-address one: geocoding is place-level (~3 km) and the grid coarsens to 1 km.
- The 'why' reasons come from a street-waterlogging relative ranking, not a probability — results/<city>_why.json labels itself that way and the card must repeat it.
- The probability is the climatology baseline over 28 cells in 2 cities unless is_learned is true; a cell not in the panel is 'not modelled', not 'safe'.
- Depth rows read 'not computable' until the HYD depth layer passes its own gate — they are not zero.
built 2026-10-14 05:41 UTC · forecast issued 2026-10-14, 7-day horizon ends 2026-10-21